Last time I covered the East; now the West is set and we can fill in the rest of the playoffs. As before, the models are described here and here. The predictions are fairly straightforward, so after that I’ll have a couple of meta-comments about the model itself.
San Antonio – Memphis: The Spurs aren’t quite as strong as their record suggest, and the Grizzlies are better than theirs. But that doesn’t mean the Grizzlies are going to upset the Spurs. San Antonio has a 84% chance of winning, which should happen in a 5 game series (it’s basically a toss-up between 5 and 6 games, both happening about a quarter of the time).
L.A. Lakers – New Orleans: The Lakers ended up with the best point differential in the West, while the Hornets have the worst (that made the playoffs). So no surprises: the Lakers win 91% of the time, 4-1.
Dallas – Portland: This will be the trendy upset pick. I’m not entirely sure when we look at the next match-up, but people are welcome to their opinions. I have the Mavericks moving on 80% of the time, probably in a 6 game series.
Oklahoma City – Denver: This should be an interesting series. Both teams have been on a tear since the trading deadline, but I think people are more optimistic about the Thunder, especially since they beat Denver twice recently. But, Denver has the better point differential for the season overall. The model says the Thunder win 49.3% of the time, so essentially a coin flip. It’ll be a six-game series either way. Going under a strict law of ‘go with the model’, I’ll take Denver in an upset.
San Antonio – Denver: The old guys against the only upset in the whole league. This will be another series where you find out how much weight people put on recent performance versus the season overall. The model just uses the season overall, so we have the Spurs 75% of the time, winning the series in 6 games. Their chances go up a little bit if OKC makes it through.
Lakers – Dallas: No one will take Dallas here, and they’re probably right. The Lakers are better than the Mavs by the same amount that the Spurs are better than the Thunder, so we have the Lakers winning 75% of the time, probably in 6 games.
San Antonio – L.A. Lakers: The Lakers are the better team, but San Antonio has home court. That’s enough to give the Spurs the edge; they win 55% of the time. Not great odds, but they should be the favorite. It’s a 6 game series.
Finals: That will leave the Bulls and the Spurs, and the Bulls got home court on the last night of the season. That will prove key, since they are also the better team and thus they have a huge advantage that the Lakers didn’t have in the conference finals. The Bulls beat the Spurs 72% of the time, and it should be a six-game series.
Now for a few meta-comments. Remember that the model only takes into account full-season point differential and home court. To the best of my knowledge, this is what everyone in the TrueHoop Smackdown uses, along with a sprinkling of their own judgment. So if someone has been traded, or someone has been hurt, the model doesn’t know. To that extent, it can be considered a kind of null hypothesis: if you can properly judge the impact of trades or injuries on a team, you should be able to adjust their strength and better evaluate how they will do in the playoffs. That being said, the model does pretty well. As described in the original post, this model is the all-time TrueHoop Smackdown champ, scoring more points in four years than any other four-year competitor. It’s in a virtual tie with Dave Berri the three years he has competed, and it’s just ahead of Justin Kubatko in his two years.
Does that mean that all these other people are doing something wrong? Not necessarily. Even across four years there are only 60 picks to be made. With the majority of points coming from just picking the right winner, there isn’t a lot of opportunity to tell different models apart. But in the long run, better evaluators should start to pull away from the model. I would argue that John Hollinger, Jeff Ma, Kevin Pelton, and Henry’s mom aren’t inspiring confidence. This year the Wages of Wins group will be having their own smackdown; we’ll see how they fair (side note: my official East picks will be there, since point differentials have changed a tiny bit in the past two days). Just keep in mind as always, predictions from one year can’t tell you who’s doing things correctly. We’ll have to check in after a few years of picks to start taking a look at who can beat the simplest reasonable model around.